Vibeleaderboard
Index / tool
Visit github.com
Category
AI Tools
Rank
No. 4035Tools index

Previous survey · No. 3760 ·

Type
TOOL
Use case
Models: Train & Run
GitHub
3 stars
Date

About

Kokoallovero is a portable CPU implementation of the Kokoro text-to-speech model, packaged as a single binary with a grapheme-to-phoneme converter, text normalizer, lexicon and guesser. It uses non-quantized SIMD code, which the author found faster than GGML/GGUF models and sometimes faster than Python torch code on their targets. It aims to simplify deploying TTS with built-in G2P instead of relying on espeak-ng.

At a glance

Kokoallovero is a standalone text-to-speech implementation around Kokoro with an English text-to-phoneme pipeline. It is worth evaluating if you want spoken output locally and need to compare CPU performance and pronunciation on your own phrases.

Good fit

Developers adding English speech to a local tool or agent.

Consider something else when

Use a multilingual engine if your input regularly mixes languages.

Start here

Prepare a synthetic sample with names, numbers and ambiguous words. Compare pronunciation and time to output on your target CPU before using it for longer passages.

Cost and requirements

Apache 2.0 for the project. A source build uses CMake. Check the model and voice assets needed for the build you choose.

Limits to know

The maintainer describes English-only phoneme conversion and incomplete text normalization. Evaluate the text you actually need rather than extrapolating from timing examples.

Tags

ttskokorosimdg2pcpuspeech

Tech Stack

Python

Comments (0)

No comments yet

Editorially curated, with community endorsements as a secondary signal. Corrections welcome.